MOSDEF: Measurements of Balmer Decrements and the Dust

MOSDEF: Measurements of Balmer Decrements and the
Dust Attenuation Curve at High Redshift
Naveen Reddy (Sloan Fellow, UC Riverside)
Collaborators:
Mariska Kriek (UCB)
Alice Shapley (UCLA)
William Freeman (UCR)
Brian Siana (UCR)
Alison Coil (UCSD)
Bahram Mobasher (UCR)
Sedona Price (UCB)
Ryan Sanders (UCLA)
Irene Shivaei (UCR)
Modeling Galaxies through Cosmic Time;
IoA, Cambridge, UK, 17 Sept 2015
Calzetti (2011)
Importance of the Dust “Curve” for High-z Galaxies
Important input to
SED fitting
Needed to infer
dust-corrected SFRs
Encodes info on the
dust/stars geometry
…combining UV and optical diagnostics of HII regions
Proxies for Dust at High-z
• UV
Slope: sensitive to age, metallicity, and star-formation
history; measurement can be complicated by presence of 2175 Å
absorption feature
• Far-IR Measurements: only available for more luminous and
dusty galaxies at high redshift (small samples of lensed galaxies)
è need tracers that are less sensitive to stellar population
parameters (age and star-formation history), probe star formation
on short timescales, and can be measured for individual typical
star-forming galaxies at high redshift
BALMER DECREMENTS
(e.g., Calzetti et al. 1994, Kennicutt et al. 2009,
Groves et al. 2012, etc…)
MOSFIRE Deep Evolution Field (MOSDEF) Survey
-Conducted using MOSFIRE on Keck (47 nights)
- MOS near-IR spectroscopy covering important
nebular emission lines at 1.4<z<3.8
- H-band-selected
Transformative
survey:
(1) large sample
of objects
(~1500)
spanning full
range of galaxy
properties
(2) multiple
redshifts to
enable
evolutionary
studies
Kriek et al. (2015)
Sampling of “Typical” (L*) Star-Forming Galaxies at z~2
Reddy & Steidel (2009)
Shivaei et al. (2015)
MOSDEF Fields/Spectra
Balmer Decrement Measurements
224 star-forming galaxies
at zspec= 1.36 – 2.59
Calculating the Attenuation Curve…
Ratios of Composites
Calculating the Attenuation Curve…
Normalization (RV)
Renormalized so that
fQeff(λè2.85 µm)=0
Normalized so that
f[Qeff(B)-Qeff(B)] = 1
Systematic
uncertainties of
ΔRV ≈ 0.4
Comparison to other common curves
Similar in shape (and normalization) to SMC at λ>2500 Å
Similar in shape (but lower normalization) than Calzetti at
λ<2500 Å
Implications for SFR(SED) and M*
≈20% lower SFRs with new curve
Δlog(M*/M¤) = 0.16 dex
Color Excesses of the Ionized Gas vs. Stellar Continuum
Price et al. (2014)
Higher attenuation
towards lines-ofsight to massive
stars
(e.g., Fanelli et al. 1988, Calzetti et al. 1994, Mas-Hesse & Kunth 1999, Kreckel et al. 2013)
Color Excesses of the Ionized Gas vs. Stellar Continuum
Assumes Cardelli+89 (Galactic) extinction curve
A Possible Physical Interpretation
At high-z: stars of all masses are attenuated by
same amount, with larger contribution of dustenshrouded SF at higher SFRs
“Low”
SFR
dominates the
UV/optical
continuum
“High”
SFR
dominates the nebular line/bolometric
luminosities
Implications for SFRs from the UV or SED-fitting
UV/SED-based SFRs underpredict total
SFR above ≈ 20 M¤/yr
Conclusions
• Large sample of Balmer decrements aids in calculating the attenuation
curve relevant for the stellar continuum
• Attenuation curve found here is similar to SMC at longer wavelengths
(λ>2500 Å), and similar in shape, but with different normalization, than
Calzetti+00
• New curve implies SFR ≈20% lower, and log M* that are 0.16 dex lower,
than those obtained with the Calzetti relation
• Difference in the color excess (and total attenuation) of the ionized gas and
stellar continuum correlates strongly with sSFR and SFR, with higher SFR
galaxies exhibiting the largest differences
• Data suggest a physical interpretation where galaxies consist of moderately
reddened stellar population that dominated the UV through near-IR
continuum, and a second, dustier population, that begins to dominate the line
and bolometric luminosities at higher SFRs.
Reddy et al. 2015, ApJ, 806, 259
Extra Slides
Recent High-Z Constraints on the Dust Curve
•
•
•
•
Noll+09
Buat+11,12
Kriek & Conroy 2013
Scoville+15
z=0.5-2.0
Kriek & Conroy (2013)
Based on photometry,
spectroscopy (in UV/optical),
and/or comparison to stellar
templates
z=2.0-6.5
Scoville+15
Implications for SFR(SED) and M*
≈20% lower SFRs with new curve
Δlog(M*/M¤) = 0.16 dex
A Possible Physical Interpretation
Locally…ionizing stars found in parent birth
clouds
Reddy et al. (2010)
Reddy et al. (2010)
Similar “Saturation” seen with IR vs UV-based SFRs
Reddy+10
Reddy et al. (2010)
Reddy et al. (2010)
Similar “Saturation” seen with IR vs UV-based SFRs
Saturation of UV
luminosity around
L*(UV)
Future Work
• Incorporate mid- and far-IR data
•Larger sample will enable studies of stellar
attenuation curve as a function of other galaxy
properties (e.g., SFR)
• Relationship between attenuation curve
shape/normalization and resolved color maps
• Multiple Balmer emission lines
MOSDEF Fields/Spectra
Balmer Decrement Measurements
224 star-forming galaxies
at zspec= 1.36 – 2.59
“Mild” Correlation
between UV slope
and τ b
Calculating the Attenuation Curve
Ratios of Composites
Spectral Shape
Limit to Galaxies of Similar Spectral Shapes
Dustinessè
Effects of Star Formation History
-“sequence” of β vs. τb
with sSFR
- are A stars contributing
to near-UV flux?
unlikely…
Effects of Metallicity?
Range of metallicity implies Δβint ≈ 0.2
Slit Losses
Dependence of the Difference
in Color Excess on SFR
Dependence of the
Difference in Total
Attenuation on SFR
Excess UV Absorption at 2175 Å?
Marginal (3σ) significance
Implications
“Low”
SFR
dominates the
UV/optical
continuum
“High”
SFR
dominates the nebular line/bolometric
luminosities
SFR(SED) and SFR(UV) may underpredict total
SFR at even “modest” levels
Appropriate attenuation curve to use for HII
regions? Gray at low SFR, MW/SMC at high SFR?